Biologicals 33 (2005) 87e94

www.elsevier.com/locate/biologicals

TSE clearance during plasma products separationprocess by Gradiflow�

Kailing Wanga,*, Annalese Johnsona, Mina Obradovica, Graeme Andersonb,Christine MacLeanb, Hari Naira

aCommercial Separations Division, Gradipore, NSW 2086, AustraliabBioReliance Corporation, Glasgow, UK

Received 24 August 2004; accepted 19 January 2005

Abstract

Background and objectives: Recent experimental evidence from rodent models suggests a potential risk for transmissible spongiformencephalopathy (TSE) transmission by blood. The emergence of a new variant CreutzfeldteJakob disease (vCJD) has raised

increased concerns about the safety of blood components and plasma products derived from vCJD-infected donors. Recent risk-minimisation strategies have included a ban on the use of UK-sourced plasma for the preparation of licensed blood products andleukodepletion of blood donations for fear of possible transmission of the human TSE via blood or blood components. The aim of

this study was to investigate the capability and efficacy of a preparative electrophoresis system (Gradiflow) in the removal of TSEcontaminants during the separation of plasma products.Materials and methods: Using hamster adapted scrapie 263K as a model for TSE agent, albumin and IgG separation from human

plasma by Gradiflow were performed separately by spiking a 263K scrapie microsomal fraction to the feed material at each processstep. Samples from pre- and post-Gradiflow separation process were titrated to the end-point for the detection of the disease-associated, proteinase K resistant form of the pathogenic prion protein (PrPSc) by Western blot.

Results: Under all conditions tested, a greater than 3 log10 reduction was achieved with no PrPSc detected in any of the pooledproducts for either of the IgG or albumin separations. These data show that Gradiflow processing has clear advantages forconcurrent purification of plasma products and in-process TSE removal.Conclusions: Our findings suggest that Gradiflow process is a viable alternative to remove causative TSE agents during plasma

products separation, potentially eliminating the risk of TSE agents transmission.� 2005 The International Association for Biologicals. Published by Elsevier Ltd. All rights reserved.

1. Introduction

Prion diseases have recently become a focus ofintense research interest, especially in Europe and theUS. Lack of understanding of the unique mechanism ofconversion from the PrPc to PrPSc isoforms and theability of related prion diseases to transmit betweenspecies have contributed significantly to this interest.

* Corresponding author. Tel.: C61 2 8977 9000; fax: C61 2 8977

9098.

E-mail address: Kailing.wang@life-therapeutics.com (K. Wang).

1045-1056/05/$30.00 � 2005 The International Association for Biologicals.

doi:10.1016/j.biologicals.2005.01.002

Whilst there is no epidemiological evidence yet tosupport CJD transmission by human blood or de-rivative products, a related disease in TSE animals hashighlighted that whole blood and its components suchas plasma and buffy coat, are capable of transmitting thedisease [1]. Classical CJD has been shown to be trans-missible by cross contamination of high titre tissuesassociated with the CNS, however, no evidence exists forits transfer via blood transfusion or fractionated products[2]. Variant CJD appears to differ from classicalCJD in that the infectious agent can be detected in thetonsils of infected patients, and thus may be present as

Published by Elsevier Ltd. All rights reserved.

88 K. Wang et al. / Biologicals 33 (2005) 87e94

a contaminant of white blood cells, and potentially bloodtransfusions.

Whilst there is no conclusive evidence of iatrogenictransmission of vCJD via blood transfusion, studies haveshown that this possibility cannot be ruled out [2]. Forthis reason regulatory authorities around the world stillconsider potentially prion contaminated biological prod-ucts as a theoretical risk. Recent risk-minimisationstrategies have included a ban on the use of UK-sourcedplasma for the preparation of licensed blood products(e.g. coagulation factors and albumin) and leukodeple-tion of blood donations. It is documented that one of themost effective methods for inactivation of TSE’s isexposure to 1 M sodium hydroxide coupled with auto-claving [3], however, such an inactivation step is entirelyunsuitable for blood derived biological products. Inaddition, methods commonly used to inactivate viral orbacterial pathogens such as autoclaving, detergent treat-ments, high temperatures or even combinations of thesemethods, do not guarantee complete inactivation of TSEagents. Therefore, the reduction of any TSE-associatedrisk will be dependent on the physical removal ofinfective TSE material during product manufacture.

Gradiflow is a membrane-based, preparative electro-phoresis technique that isolates proteins by exploiting theintrinsic characteristics of molecular weight and charge.The apparatus employs polyacrylamide membranes withdefined pore sizes enabling separation of molecules basedon size, whilst the pHof the electrophoresis buffer chargesproteins depending on their isoelectric points [4]. The useof mild buffer conditions and effective heat dissipationminimises protein denaturation and enables the processto be scaled up. The selection of the relevant electro-phoretic separation membrane and buffer enables thepurification of molecules and simultaneous removal ofbiological contaminants such as viruses and endotoxins.Importantly, this technology is fully scalable, makingthis platform useful for integration into plant scalemanufacturing processes. The large scale purification ofimmunoglobulinG fromhumanplasmabyGradiflowhashighlighted the advantages of separation process in termsof consistently highproduct yield, purity and safety [5].Asa result, Gradiflow represents a viable alternative toconventional plasma fractionation and precipitationprocesses.

To evaluate the efficacy of removing TSE contami-nants by Gradiflow during plasma products separationprocess, a TSE clearance study was conducted atBioreliance Corporation (formerly Q-One Biotech Ltd,UK). The standard validation protocol for Biorelianceutilised hamster adapted scrapie 263K as amodel for TSEagent. Albumin and immunoglobulin G (IgG) separa-tions by Gradiflow from human plasma were performedseparately by spiking a 263K scrapie microsomal fractionto the feed material at each process step. Samples frompre- and post-Gradiflow processes were titrated to the

end-point for the detection of the disease-associated,proteinase K resistant form of the pathogenic prionprotein (PrPSc) by Western blot. Whilst bioassays are themost reliable method of assaying for the presence ofinfectious prion, Western blotting was chosen for itsrelative speed and reduced costs for preliminary valida-tion work. The result of the Bioreliance protocol forWestern blotting was a semi-quantitative blot thatdemonstrated Gradiflow can be used as an effective toolto remove causative TSE agents.

2. Methods

2.1. Gradiflow apparatus

For an overview of the Gradiflow instrument seereference [6]. Briefly, the essential element of the Gradi-flow technology is the separation cartridge, which consistsof three polyacrylamide membranes. The separationmembrane is located between two restrictionmembranes,thereby partitioning stream 1 (S1) and stream 2 (S2).Typically, feed material is loaded into S1 (feed stream)and target proteins are transferred across the separationmembrane to S2 (product stream) when an electricpotential is applied. The movement of proteins dependson the pore size of the membranes, the pH of electropho-resis buffer and the configuration of the electrodes. Inthese experiments, small pore size membranes enclose S1and S2, preventing contact of the infected prion with theelectrodes of the separation unit. This is important asprion has the potential to bind to metal surfaces, whichcould result in cross contamination of experiments [7]. Allstreams are kept circulating and chilled to dissipate theheat produced in the electrophoretic process.

All scrapie partitioning experiments were performedusing a PF100 (Gradipore Ltd, Sydney), a preparativeprotein electrophoresis instrument adapted for pathogenvalidation study. The PF100 operates on an identicalprinciple to the research and pilot scale versions of thistechnology, the difference being this machine is fullyautoclavable and stream 1 (S1) and stream 2 (S2) aresealed for pathogen containment. In all scales, thecartridge is disposable, therefore fresh membranes andhousing as well as stream tubing were utilised for eachexperiment, removing risk of prion transmission. At thecompletion of the run, the apparatus was decontaminatedby soaking with 2 M NaOH for at least 60 min followedbyneutralisationwith 1 MHCl.ResidualNaOHandHClwere removed by three washes with H2O. The apparatuswas autoclaved prior to reuse.

2.2. Preparation of microsomal scrapie

The hamster adapted scrapie strain 263K was used asa model for potential TSE contaminants. Scrapie brain

89K. Wang et al. / Biologicals 33 (2005) 87e94

homogenate was prepared and aliquotted by BiorelianceCorporation. Brains from 263K scrapie infected ham-sters were homogenised in PBS, to a final concentrationof 10% (w/v). The homogenate was clarified by lowspeed centrifugation, to remove larger cell debris and thesupernatant was collected. The supernatant material wasthen clarified again by centrifugation at 10 000 g for10 min, before being ultracentrifuged at 141 000 g for60 min, to concentrate the scrapie fibrils, small mem-brane vesicles and fragments. The pelleted material ethe microsomal fraction e was then resuspended in PBS,aliquotted and stored at �70 �C.

2.3. Gradiflow separation process for IgG andalbumin

Frozen normal pooled plasma (obtained from ScottishNational Blood Transfusion Service, UK) was thawed at37 �C. For IgG purification, 5 ml of plasma was dilutedwith 10 ml of Mes-Bis Tris buffer, pH 5.4 (41 mM 2-[N-morpholino]ethanesulfonic acid, 6.3 mM bis(2-hydrox-yethyl) imino-tris(hydroxymethyl) methane) and loadedinto S1. A 263Kmicrosomal fractionwas added into S1 ata volume ratio of 1:10.Amembrane cartridge, comprisingan IgG transfer specific separation membrane sand-wiched between two small pore size restriction mem-branes was used in the separation unit. An electricpotential of 250 V, with the positive electrode configuredat S1 was applied across the separation unit. Electropho-resis separations were performed using aMes-Bis Tris pH5.4 buffer.

For albumin purification, 5 ml of plasma was dilutedwith 10 ml of TriseBorate buffer, pH 9.0 (32 mMTrizmabase, 96 mM Boric acid) and loaded into S1. A 263Kmicrosomal fraction was added into S1 at a volume ratioof 1:10. A membrane cartridge, comprising an albumintransfer specific separation membrane sandwiched be-tween two small pore size restrictionmembranes was usedin the separation unit. An electric potential of 250 V, withthe negative electrode configured at S1 was applied acrossthe separation unit. Electrophoresis separations wereperformed using TriseBorate buffer pH 9.0.

The product in S2 was harvested at 60, 120, 180, 240and 300 min intervals. After each harvest, 10 ml of bufferwas used to replenish S2. Samples from both streams (S1and S2) pre- and post-separation process were collectedfor PrPSc titration by Western blot assay.

2.4. Ultracentrifugation of samples

Samples were thawed at 37 �C and sonicated on ice tobreak up aggregates of the prion. The sample wasaliquotted to a labelled ultracentrifuge tube and thevolume topped up to 39 ml with PBS as required. Thesamples were centrifuged at 28 000 rpm for one hour at4 �C. After centrifugation, the supernatant was decanted

from the tube and the pellet allowed to drain. The pelletwas resuspended in an appropriate volume of PBSaccording to concentration desired. Samples were thenaliquotted, labelled and stored at �70 �C until required.

2.5. Western blotting assay

A Western blot assay (developed and validated byBioreliance) was used to detect PrPSc in the samplesgenerated by PF100 processing. Briefly, samples wereneutralised and ultracentrifuged to concentrate the PrPSc.The pellets were resuspended in PBS, sonicated anddigested with Proteinase K (see Optimisation of PKDigest). Serial 5-fold dilutions of the samples in SDSdenaturing buffer (10% SDS, 50 mM Tris/HCl contain-ing b-mercaptoethanol) were prepared. The sampledilutions were separated by electrophoresis on 12%polyacrylamide gels (BioRad) at 150 V for 45 min. Thegel was blotted onto a polyvinylidene difluoride (PVDF)membrane (Imobilin-P, Millipore) using a semi-drytransfer apparatus (BioRad) at 15 V for 1 h. Themembranes were blocked in 5% skim milk (Marvel)and PBST (PBS and 0.05% Tween 20) at 4 �C overnight.The membranes were probed with 3F4 primary antibody(Senetek PLC) followed by sheep anti-mouse IgG HRPconjugated secondary antibody. The signal was detectedusing ECL Plus� (Amersham). Films were developedusing an automated X-ray developer machine.

2.6. Optimisation of proteinase K digestion

Process samples often contain proteins which caninterfere with proteinase K digestion, or which caninterfere with the detection of 263K PrPSc proteins inthe Western blot assay. Thus, prior to performing thespiked process runs, initial studies were conducted tooptimise the concentration of proteinase K required toremove background proteins while allowing effectivedigestion of cellular prion protein (PrPc).

A set of control samples (S10, S1end, S2end) from anunspiked IgG separation run were ultracentrifuged,resuspended in PBS and digested at 37 �C with pro-teinase K at final concentrations of 3.16, 10, 31.6, 100,316, 1000, and 3160 mg/ml for 1 h. The samples wereanalysed by Western blotting.

To confirm that digestion with proteinase K wouldallow detection of PrPSc at low concentrations, thecontrol samples from the unspiked IgG separation werespiked with 263K microsomal fraction stock at a finaldilution of 1/125 (10�2.1). The samples were digestedunder the conditions established for the unspikedproteinase K digest. Three serial 5-fold dilutions ofeach digested sample were analysed by Western blotting,alongside a positive control of the 263K stock.

90 K. Wang et al. / Biologicals 33 (2005) 87e94

2.7. Calculation of PrPSc reduction factors

The input spike of a sample was calculated bydetermining the titre of the straight microsomal PrPSc

based on Western blotting. The titre was taken as thereciprocal of the first dilution at which no PrPSc could beseen. The total level of PrPSc in the starting material wascalculated by multiplying the titre obtained (in arbitraryunits/ml) by the sample volume and also by necessarycorrection factors such as dilution due to pH neutralisa-tion, or concentration due to ultracentrifugation. Resultswere subsequently expressed in terms of reduction factor(RF) where:

RFZTotal input of PrPSc in stream 1

�feed stream

�

Total output of PrPSc in stream 2�product stream

�

The calculated RF values are expressed in thelogarithmic (log10) form. If no PrPSc was detected inthe undiluted sample, then the reciprocal of the dilutionwas assumed to be 1 and the RF values are givenasR the calculated log10 value.

3. Results

3.1. Removal of PrPSc from plasma products IgGand albumin

3.1.1. IgG runsFor these runs, the load sample consisted of 1/10

dilution of scrapie and 1/3 dilution of plasma in thedesignated running buffer. All spiked process sampleswere ultracentrifuged and subsequently analysed byWestern blotting. Once again, a straight positive controland an ultracentrifuged positive control were includedto validate the blots. Digested straight positive controlwas also included in each of the individual sample blots.

Several IgG separation runs were performed underidentical conditions. Due to the nature of the analysis,representative sets of data are shown for each set ofruns. A representative set of blots for the IgG separationruns are shown in Fig. 1. Total PrPSc within each sampleand the final reduction factors were calculated asdescribed in the methods (calculation of reductionfactors). The representative results are summarised inTable 1.

Investigative studies were performed to re-optimisethe proteinase K concentration used to digest S2 samples.263K recovery experiments were performed in thepresence of 10e316 mg/ml proteinase K, and there wasno apparent loss of the PrPSc protein at any conditiontested. Thus, digestion of the S2 sample under theseconditions resulted in removal of the protein bandsdetected in the sample, but did not result in removal of

extraneously added PrPSc. Whilst ensuring there was noloss of PrPSc due to over-digestion, PrPSc bands around28 kDa were absent in each of S2 samples tested,suggesting the effective removal of PrPSc by this process.

Western blot analysis was performed on one of thestream 2 samples in the absence of primary antibody.The background bands (including the 28 kDa species)were still detectable, although slightly reduced inintensity, when blotting was performed using secondaryantibody alone. In addition, the 263K marker was notvisible in the absence of primary antibody. These resultssuggest that most of the background in the S2 samplesresults from cross-reactivity, or non-specific reactivity,of partially digested Ig with the secondary antibody. Itshould be noted that Ig light chains migrate on SDS-PAGE with a similar MW to the 28 kDa PrPSc protein,and may cross-react with the secondary antibodies. Theoccurrence of this partially digested Ig in the blotssuggests that the microsomal spike may have aninhibitory effect on the Ig digestion by PK.

There was a higher level of background in stream 2plasma samples from the spiked process runs, than inthe control sample tested and it was therefore notpossible to identify the 28 kDa species in stream 2samples as PrPSc. Digestion of a second, unspiked S2control with different concentrations of proteinase Kgave similar results to the initial experiments (the samplewas clean using a final concentration of 10 mg/mlproteinase K), suggesting that no problem had arisenduring the generation of the original control sample.

3.1.2. Albumin runsSeveral albumin separation runs were performed

under identical conditions. A representative set of blotsfor the albumin separation runs are shown in Fig. 2.Total PrPSc within each sample and the final reductionfactors were calculated as described in the methods(calculation of reduction factors). These results aresummarised in Table 2.

The results of the IgG and albumin separation runsboth show complete partitioning of PrPSc from thedesired protein product (IgG or albumin) withinthe limits of detection of the Western blot technique.The majority of the spiked PrPSc is recovered in stream 1of the IgG and albumin separation run with a reductionfactor of 0.2 and 0.3, respectively, while no PrPSc isdetected in stream 2 (product stream) with a reductionfactor of 3.3 and greater than (or equal to) 2.6 logs,respectively.

The Western blot results of the digested positivecontrol (present in the blots of all the processed samples)show a broad band at w26e28 kDa; a broad, but lessintense band at w23e24 kDa, and a faint, sharper bandat approximately 19 kDa. The 19 kDa species is thoughtto represent the non-glycosylated protein. In an un-digested sample containing PrPSc, the precursor PrP

91K. Wang et al. / Biologicals 33 (2005) 87e94

A B C

D E

1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10

1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10

Fig. 1. Titration of run samples from IgG separation. A: S1 starting material. B: S1 end material. C: S2 end material. D: positive control (input

spike). E: S2 end material with uninfected spike material. Lane 1: 10�2.1 263K marker; lane 2: MW marker; lane 3: undigested sample, neat dilution;

lanes 4e10: sample digested with proteinase K to a final concentration of 10 mg/ml and titrated from neat dilution to 10�4.2 dilution. For E: lane 1:

10�2.1 263K marker; lane 2: MW marker; lane 3: undigested sample, neat dilution; lane 4: neat sample digested with proteinase K at a final

concentration of 100 mg/ml; lanes 5e10: sample digested with proteinase K to a final concentration of 100 mg/ml and titrated from neat dilution to

10�3.5 dilution.

protein had an apparent molecular weight of around33e35 kDa. An additional w54 kDa species is alsooccasionally observed.

4. Discussion

In this study, hamster adapted scrapie 263K was usedas a model agent to assess TSE clearance in Gradiflowseparation process leading to the isolation of plasmaderivatives such as IgG and albumin. Different TSEspiking agents have different properties, for examplemembrane bound prion spiking agents will partitiondifferently from unbound pathogenic protein prepara-tions [8]. This makes it advisable to employ a spiking

Table 1

IgG separation

Sample Total

PrPSc

(arbitrary units)

Reduction Reduction

(log10)

Positive control 3.4! 104 e e

S1 (load sample) 1.3! 104 e e

S1 (end material) 2.1! 104 6.2! 10�1 eS2 (end material) 7.0! 100 1.9! 103 3.3a

a This S2 sample was subsequently re-analysed on a Western blot

after re-optimisation of proteinase K digestion to indicate complete

absence of PrPSc from S2 (see Section 4).

agent which has suitable properties to mimic thepartitioning of the theoretical contaminant [8]. In thisstudy we have utilised hamster adapted scrapie whichexhibits similar partitioning behaviour to human vCJD(PrPvCJD) [9].

All samples generated during this study were testedfor the detection of the disease-associated, proteaseresistant form of the hamster prion protein (PrPSc) usinga Western blot assay. Although the bioassay remains theonly method by which to measure infectivity, levels ofPrPSc detected by Western Blot have been shown tocorrelate closely with the level of infectivity detectable bybioassay [13]. Thus, theWestern blot can be used in theseinitial studies to provide an early indication of theeffectiveness of process steps to remove the scrapie agent.

Whilst epidemiological evidence does not suggest thatCJD is or has been transmitted via blood transfusion,a recent fatal case of vCJD in the UK arising fromtransfusion from a donor who in turn died of vCJDsuggest the alarming possibility that the infection may betransfusion transmissible [2]. The concern now is not onlyfor whole blood transfusions, but also plasma derivedproducts including coagulation factor concentrates,immunoglobulin preparations, albumins and fibrin seal-ants. In the past, a number of such products have hadrecall notices issued following the identification of CJD

92 K. Wang et al. / Biologicals 33 (2005) 87e94

A CB

1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10

Fig. 2. Titration of run samples from albumin separation. A: positive control (input spike). B: S1 end material. C: S2 end material. Lane 1: 10�2.1

263K marker; lane 2: MWmarker; lane 3: undigested sample, neat dilution; lanes 4e10: sample digested with proteinase K to a final concentration of

10 mg/ml and titrated from neat dilution to 10�4.2 dilution.

positive donors in the plasma pools used for production,and it seems likely that regulatory agencies will continuethese bans until such time as the risks posed by bloodderived products can be assessed more quantitatively.Therefore, appropriate controls during manufacture andvalidation of manufacturing process for prion clearanceappear to be the viable alternative in the absence of analternate strategy.

The level of concern surrounding potential contami-nation of therapeutic products with prion, coupled withthe difficulty in destroying the agents of TSE disease hasopened a new market to the TSE removal processes.Whilstmethods are available for simultaneous removal ofTSE and fractionation of plasma proteins, such as Cohnfractionation, most methods involving removal of theTSE rely on size exclusion filtration [10,11], or depthfiltration [12] and are performed in addition to an existingfractionation process. This study has looked at theefficacy of the Gradiflow� technology as a tool forplasma products purification and in-process simulta-neous prion removal. This technology, operating on theprinciples of membrane-based electrophoretic fraction-ation, has clearly been shown to effectively remove PrPSc

from the target plasma proteins albumin and IgG.TheGradiflow technology exploits two primarymodes

of electrophoretic separation to isolate target proteins.Charge based separation is based on selection of a bufferwhose operating pH is between the pI of the target proteinand the pI of the contaminants. This will give the target

Table 2

Albumin separation

Sample Total

PrPSc

(arbitrary units)

Reduction Reduction

(log10)

Positive control 2.9! 104 e eS1 (load sample) 4.9! 102 e e

S1 (end material) 1.4! 104 3.5! 10�2 e

S2 (end material) 1.3! 100 3.8! 102 R2.6

and the contaminants a differential charge, allowingpartitioning in the separation unit.

Size based separation utilises a buffer of a high pH toimpart a like charge on all proteins in the field. Theseparation membrane pore size used is then selective forpermeation of the target protein away from the contam-inants. Alternatively, a membrane can be selected toallow transfer of contaminants away from the targetprotein.

The work performed in this study incorporated boththe size and charge based separations simultaneously toachieve isolation of the target plasma proteins. Separa-tion membranes used were designed specifically for thetransfer of the nominated protein, albumin or IgG,respectively, with small pore sized separations adopted tosandwich the streams and prevent movement of prion orthe target protein out of the streams. The separationmembranes not only prevent transfer of the contaminantscrapie protein, but also themovement of large cell debristo which some prion protein may be associated. Theoverall conditions (size and charge) selected for thealbumin and IgG separations create an environmentconducive to not only the movement of the targetedplasma protein to the product stream, but also theretention of the spiked prion protein in the feed stream.

Results given in this study are representative of theresults observed across an entire set of experiments. Dueto the nature of Western blotting, it is impossible togenerate an average data set from a semi-quantitativeassay. The Western blotting assay used by Bioreliance intheir validation work has been internally validated. Thecorrelation of results by Western blotting and those bybioassay has been demonstrated previously by Lee et al.[13]. The Western blot results of all the IgG separationruns performed showed good recovery of PrPSc in samplestream 1 at the completion of the run (Fig. 1B erepresentative blot), with total PrPSc levels similar to orwithin 5-fold of those seen in the corresponding loadsamples. The end-point of stream 2 samples was moredifficult to interpret, as there were a number of bands

93K. Wang et al. / Biologicals 33 (2005) 87e94

present in the S2 samples, including a band at 28 kDa.This background was not seen in the unspiked S2 controlsample (data not shown).

The control samples used in preliminary studies weregenerated from an unspiked plasma run performed underidentical process conditions to the spiked plasma IgGseparation runs. It was, therefore, considered possiblethat the presence of the microsomal fraction within thespiked runs resulted in the precipitation of a higher levelof contaminants during the ultracentrifugation step, andthat unspiked control samples may not be representativeof the stream2 samples from the spiked process runs.Newcontrol samples were thus generated from a Gradiflowrun using an uninfected hamster microsomal fraction asthe spiking agent, and ultracentrifuged as before. Back-ground in this stream 2 sample was higher than thatobserved for the original control, and the final concen-tration of 100 mg/ml proteinaseKwas required to removethe background, 28 kDa species (presumably Ig) asshown in Fig. 1E. Therefore, the outcome of this investi-gation demonstrated that the band is not a result of PrPSc

being present in the sample, but an artefact of the runitself.

As with the IgG runs, no PrPSc was detected in stream2 samples of any of the albumin separation runs.Therefore, complete partitioning of PrPSc from stream 2in the albumin runs was achieved, within the limits ofdetection of the Western blot assay, giving reductionvalues of R2.6 logs (Table 2) and R1.8 logs (data notshown). The Western blot results of all the albuminseparation runs performed showed good recovery ofPrPSc in sample stream 1 with regard to the input spike,although recovery in two separate runs differed by 1 log.There was considerable loss of PrPSc in the load samples,with drops of 1.8 and 2.5 logs. This was a phenomenonobserved occasionally throughout the experimental dataset, and was determined by Bioreliance to be the result ofPrPSc loss during ultracentrifugation. Ultracentrifuga-tion of PrPSc usually results in either no reduction, or upto 5-fold reduction in the level of PrPSc. Each set ofprocess samples was centrifuged alongside a positivecontrol, and on each occasion there was good recovery ofthe PrPSc protein with titres either identical to or at most5-fold less than the straight positive control. However, asthe load sample is not subjected to Gradiflow processingbefore sampling, the inconsistency in recovery in the loadstream must arise from the ultracentrifugation processand the limits of the Western blotting sensitivity.

5. Conclusions

The purpose of this study was to investigate thepotential application of the Gradiflow� system forremoval of 263K scrapie agent. A number of processconditions tested resulted in significant partitioning of

PrPSc from buffer and protein samples, thus demonstrat-ing effective prion clearance from target proteins. Theoutcomes of the spiked runs using two different sets ofrunning conditions presented similar results demonstrat-ing that the majority of the PrPSc was recovered in stream1 with no PrPSc being detectable in any of the suggestedproduct stream (S2) samples tested. This demonstratesthe effectiveness of the separation protocol in partitioningPrPSc in the Gradiflow�.

Whilst the plasma IgG separation runs were slightlymore difficult to interpret, again the majority of PrPSc

was recovered in stream 1, in all processed samples.Investigative studies proved the presence of additionalbackground bands to be due to cross-reactivity of thesecondary antibody with Ig proteins that may poten-tially co-migrate with PrPSc. While it is difficult tounequivocally confirm that the presence of PrPSc wasnot being masked by this protein, further proteinase Koptimisation studies demonstrated the band could becompletely removed from the sample without loss ofPrPSc, making the presence of PrPSc highly unlikely.

The albumin runs also demonstrated complete parti-tioning of the PrPSc from the S2 samples, with highrecovery of PrPSc observed in the load stream samples.These results have shown that the Gradiflow� system iscapable of removing PrPSc from human plasma compo-nents, IgG and albumin and would further suggest theGradiflow system has the potential to separate the scrapieprion protein from other plasma components.

PrPSc has been used as a marker for infectivity,however, due to the limited sensitivity of the blottingtechnique, effective removal of the infectious agentawaits the use of bioassay for final confirmation. Ourfindings suggest that Gradiflow has the potential toremove infectious TSE agents during plasma productspurification, thus eliminating the risk of prion diseasetransmission.

References

[1] Houston F, Foster JD, Chong A, Hunter N, Bostock CJ.

Transmission of BSE by blood transfusion in sheep. Lancet

2000;356:999e1000.

[2] Llewelyn CA, Hewitt PE, Knight RSG, Amar K, Cousens S,

Mackenzie J, et al. Possible transmission of variant CreutzfeldteJakob disease by blood transfusion. Lancet 2004;363:417e21.

[3] Taylor DM. Inactivation of TSE agents: safety of blood and

blood derived products. Transfus Clin Biol 2003;10:23e5.

[4] Rylatt DB, Napoli M, Ogle D, Gilbert A, Lim S, Nair CH.

Electrophoretic transfer of proteins across polyacrylamide mem-

branes. J Chromatogr A 1999;865:145e53.

[5] Li G, Stewart R, Conlan B, Gilbert A, Roeth P, Nair H.

Purification of human immunoglobulin G: a new approach to

plasma fractionation. Vox Sang 2002;83:332e8.

[6] Thomas TM, Shave EE, Bate IM, Gee SC, Franklin S, Rylatt DB.

Preparative electrophoresis: a general method for the purification

of polyclonal antibodies. J Chromatogr A 2002;944:161e8.

94 K. Wang et al. / Biologicals 33 (2005) 87e94

[7] Weissman C, Enari M, Klohn P-C, Rossi D, Fleschig E.

Transmission of prions. JID 2002;186:S157e65.

[8] Vey M, Baron H, Weimer T, Groner A. Purity of spiking agent

affects partitioning of prions in plasma protein purification.

Biologicals 2002;30(3):187.

[9] Stenland CJ, Brown P, Lee DC, Petteway Jr SR, Rubenstein R.

Partitioning of human and sheep forms of the pathogenic prion

protein during the purification of therapeutic proteins from human

plasma. Transfusion 2002;42:1497e500.

[10] Tateishi J, Kitamoto T, Mohri S, Satoh S, Sato T, Shepherd A,

et al. Scrapie removal using Planova virus removal filters.

Biologicals 2001;29(1):17e25.

[11] Van Holten RW, Autenrieth S, Boose JA, Hsieh WT,

Dolan S. Removal of prion challenge from an immune globulin

preparation by use of a size exclusion filter. Transfusion

2002;42(8):999e1004.[12] Van Holten RW, Autenrieth SM. Evaluation of depth filtration to

remove prion challenge from an immune globulin preparation.

Vox Sang 2003;85(1):20e4.

[13] Lee C, Stenland CJ, Hartwell RC, Ford EK, Cai K, Miller JL,

et al. Monitoring plasma processing steps with a sensitive

Western blot assay for the detection of the prion protein. J Virol

Methods 2000;84(1):77e89.